Method and system for non-contact identification of nine types of constitutions in traditional Chinese medicine

Abstract

The embodiment of the invention discloses a method and system for non-contact identification of nine types of constitutions in traditional Chinese medicine, and the method comprises the steps: collecting N face videos, where N is an integer greater than 10; the face video is processed through a remote photoelectric volume pulse wave tracing method, and a remote photoelectric volume pulse wave signal is obtained; carrying out heart rate variability characteristic analysis on the remote photoelectric volume pulse wave signal to obtain a heart rate variability characteristic; constructing an identification model, and inputting the heart rate variability characteristics into the identification model for training to obtain an optimal identification model; and inputting a to-be-identified heart rate variability feature into the optimal identification model to obtain a traditional Chinese medicine nine-type constitution identification result corresponding to the to-be-identified heart rate variability feature. The problem that the traditional Chinese medicine constitution cannot be scientifically, accurately and conveniently predicted in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly relates to a method and system for non-contact identification of nine types of traditional Chinese medicine constitutions. Background Art

[0002] Traditional Chinese medicine constitution is a comprehensive and relatively stable trait formed by an individual in terms of morphological structure, physiological function, and psychological state under the action of congenital inheritance and acquired environment. At present, the classification and determination of traditional Chinese medicine constitutions mainly rely on the nine types of constitution scales issued by the Chinese Association of Chinese Medicine. These nine constitution types include peaceful constitution, qi-deficiency constitution, yang-deficiency constitution, yin-deficiency constitution, phlegm-dampness constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution, and special endowment constitution.

[0003] However, the existing methods for determining constitution scales have significant defects. The results completely depend on the answers to the scale questions, and the answers to the questions are divided into five levels. In actual operation, since the degrees represented by 2 points, 3 points, and 4 points are relatively close, the boundaries are blurred, which often leads to large deviations in the final conversion results. In addition, this determination method is highly subjective, and the entire operation process is also relatively cumbersome and complex, posing challenges to the accurate determination of traditional Chinese medicine constitutions.

[0004] In summary, the existing methods for classifying and determining traditional Chinese medicine constitutions have problems such as low accuracy, strong subjectivity, and complex operation, seriously affecting the effect and popularization of traditional Chinese medicine constitution classification in practical applications. There is an urgent need for a more scientific, accurate, and convenient method for predicting traditional Chinese medicine constitutions. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide a method and system for non-contact identification of nine types of traditional Chinese medicine constitutions, so as to solve the problem in the prior art that it is impossible to scientifically, accurately, and conveniently predict traditional Chinese medicine constitutions.

[0006] To achieve the above purpose, the embodiments of the present invention provide a method for non-contact identification of nine types of traditional Chinese medicine constitutions. The method for non-contact identification of nine types of traditional Chinese medicine constitutions specifically includes:

[0007] Collect N face videos, where N is an integer greater than 10;

[0008] Process the face video through remote photoplethysmography to obtain the remote photoplethysmogram signal;

[0009] Perform heart rate variability feature analysis on the remote photoplethysmogram signal to obtain heart rate variability features;

[0010] Construct an identification model, input the heart rate variability features into the identification model for training to obtain an optimal identification model;

[0011] Input the heart rate variability features to be identified into the optimal identification model to obtain the identification results of the nine major types of traditional Chinese medicine constitutions corresponding to the heart rate variability features to be identified.

[0012] Based on the above technical solutions, the present invention can also be improved as follows:

[0013] Further, the analysis of the heart rate variability features of the remote photoplethysmogram signal to obtain the heart rate variability features includes:

[0014] Perform light intensity analysis on each frame image in the face video to obtain a signal of the change of light intensity over time, and extract the original rPPG signal based on the signal;

[0015] Preprocess the original rPPG signal to obtain a pure rPPG signal.

[0016] Further, the analysis of the heart rate variability features of the remote photoplethysmogram signal to obtain the heart rate variability features further includes:

[0017] Extract the heart rate variability features from the pure rPPG signal;

[0018] Among them, the heart rate variability features include time domain feature parameters, frequency domain feature parameters, and non-linear feature parameters.

[0019] Further, the time domain feature parameters include the deviation of heart rate variability from the average value, the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals, the root mean square of adjacent RR interval differences, the standard deviation of adjacent RR interval differences, the coefficient of variation of the standard deviation of RR intervals, the measurement value of HRV size, and the degree of deviation of heart rate from the average value:

[0020] Calculate the deviation of heart rate variability from the average value through formula (1);

[0021]

[0022] In the formula, SDNN is the deviation of heart rate variability from the average value, N is the total number of cardiac cycle intervals, RR i is the i-th RR interval, is the average value of the RR intervals of N cardiac beats;

[0023] Calculate the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals through formula (2);

[0024]

[0025] Wherein, PNN50 is the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals, NN50 is the number of adjacent RR interval differences greater than 50 ms, and N is the total number of RR intervals;

[0026] Calculate the root mean square of adjacent RR interval differences through formula (3);

[0027]

[0028] Wherein, RMSSD is the root mean square of adjacent RR interval differences, N is the total number of R intervals, and RR i+1 is the (i + 1)-th RR interval;

[0029] Calculate the standard deviation of adjacent RR interval differences through formula (4);

[0030]

[0031] Wherein, RR′ i is the difference of the i-th adjacent RR interval;

[0032] Calculate the coefficient of variation of the standard deviation of RR intervals through formula (5);

[0033]

[0034] Wherein, CVSD is the coefficient of variation of the standard deviation of RR intervals, and RMSSD is the root mean square of adjacent RR interval differences;

[0035] Calculate the HRV magnitude measurement through formula (6);

[0036]

[0037] Wherein, CVnni is the HRV magnitude measurement;

[0038] Calculate the degree of deviation of heart rate from the average value through formula (7);

[0039]

[0040] Wherein, std_hr is the degree of deviation of heart rate from the average value, HR i represents the i-th previous heart rate, and HR is the heart rate.

[0041] Furthermore, the frequency domain characteristic parameter includes the ratio of low-frequency to high-frequency power spectra;

[0042] Calculate the ratio of low-frequency to high-frequency power spectra through formula (8);

[0043]

[0044] Wherein, LF is the low-frequency power spectrum, HF is the high-frequency power spectrum, f(λ) is the power spectrum density function, and dλ is the tiny change in frequency.

[0045] Furthermore, the non-linear characteristic parameters include short-term heart rate variability and long-term heart rate variability:

[0046] Calculate the short-term heart rate variability through formula (9):

[0047]

[0048] Wherein, SD1 is the short-term heart rate variability;

[0049] Calculate the long-term heart rate variability through formula (10):

[0050]

[0051] Wherein, SD2 is the long-term heart rate variability.

[0052] Furthermore, the method for non-contact identification of nine types of traditional Chinese medicine constitutions further includes:

[0053] Calculate the sample entropy through formula (11);

[0054]

[0055] Wherein, SampEn(m, r) is the sample entropy, B m (r) is the number of vector pairs with a length of m whose distance is less than or equal to r, B m+1 (r) is the number of vector pairs with a length of m + 1 whose distance is less than or equal to r, is the limit when approaching infinity;

[0056] Calculate the information entropy through formula (12);

[0057] Gain(F) = Info(s) - Info F (S) Formula (12)

[0058] Wherein, Gain(F) is the information gain of feature F, Info(S) is the information entropy of dataset S, and Info F (S) is the conditional entropy of dataset S under the condition of feature F.

[0059] A system for non-contact identification of nine types of traditional Chinese medicine constitutions includes:

[0060] An acquisition module, configured to acquire N face videos, where N is an integer greater than 10;

[0061] A preprocessing module for processing the face video through remote photoplethysmography to obtain the remote photoplethysmogram signal;

[0062] A feature analysis module for analyzing the heart rate variability features of the remote photoplethysmogram signal to obtain heart rate variability features

[0063] A construction module for constructing an identification model, inputting the heart rate variability features into the identification model for training to obtain an optimal identification model;

[0064] An identification module for inputting the heart rate variability features to be identified into the optimal identification model to obtain the traditional Chinese medicine constitution identification result corresponding to the heart rate variability features to be identified.

[0065] An electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the method are implemented.

[0066] A non-transitory computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method are implemented.

[0067] The embodiments of the present invention have the following advantages:

[0068] In the method for non-contact identification of nine major types of traditional Chinese medicine constitutions in the present invention, data is obtained by collecting face videos without direct contact with the human body. Compared with traditional methods for identifying traditional Chinese medicine constitutions, it avoids the cumbersome problem answering and subjective scoring processes when using physical scales, greatly improving the convenience of operation.

[0069] By performing HRV feature analysis on the remote photoplethysmogram signal, the identification of traditional Chinese medicine constitutions can be transformed from subjective judgment to objective quantitative data. This helps to reduce the subjectivity and uncertainty of human judgment.

[0070] There is a correlation between traditional Chinese medicine constitutions and the regulation of autonomic nerve function, and HRV can well reflect autonomic nerve function. Therefore, analyzing HRV features to identify traditional Chinese medicine constitutions has a strong physiological relevance.

[0071] For example, common sleep problems, emotional responses, and gastrointestinal symptoms in the types of biased constitutions are all related to autonomic nerve dysfunction. HRV feature analysis can provide a basis for the identification of traditional Chinese medicine constitutions from a physiological mechanism.

[0072] By constructing and training an identification model, an optimal identification model can be obtained, which can accurately classify the nine major types of traditional Chinese medicine constitutions. A large amount of face video data (N is an integer greater than 10) provides a rich data basis for model training, which helps to improve the generalization ability and identification accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, other implementation drawings can be obtained according to the provided drawings without creative efforts.

[0074] The structures, ratios, sizes, etc. shown in this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention. Therefore, they do not have technical substantive significance. Any modification of the structure, change of the proportional relationship or adjustment of the size should still fall within the scope covered by the technical content disclosed in the present invention without affecting the effects that the present invention can produce and the purposes that can be achieved.

[0075] Figure 1 It is a flowchart of the method for non-contact identification of the nine major types of traditional Chinese medicine constitutions of the present invention;

[0076] Figure 2 It is an architecture diagram of the system for non-contact identification of the nine major types of traditional Chinese medicine constitutions of the present invention;

[0077] Figure 3 It is a schematic diagram of the physical structure of the electronic device provided by the present invention.

[0078] Wherein the reference numerals are:

[0079] Acquisition module 10, preprocessing module 20, feature analysis module 30, construction module 40, identification module 50, electronic device 60, processor 601, memory 602, bus 603. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0080] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0081] It should be noted that in the description of the present invention and the above-mentioned drawings, terms such as "first" and "second" are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so as to implement the present invention described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0082] It should be noted that, without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The embodiments of the present invention will be described in detail below with reference to the drawings. Figure 1 The following is a flowchart of an embodiment of the method for non-contact identification of nine types of traditional Chinese medicine constitutions according to the present invention. As Figure 1 shown, a method for non-contact identification of nine types of traditional Chinese medicine constitutions provided by an embodiment of the present invention includes the following steps:

[0083] S101, collect N face videos, where N is an integer greater than 10;

[0084] Specifically: Use a high-definition camera as the acquisition device to ensure that its resolution can clearly capture the subtle changes in the face. For example, the resolution is at least 1920×1080 pixels. Install the camera at a fixed position to ensure that the shooting angle can completely capture the face image. Generally, it is recommended to keep it horizontal with the face of the person being collected, and the distance is between 0.5-1.5 meters.

[0085] The acquisition environment should maintain uniform and sufficient light, avoiding direct sunlight or shadow interference. The combination of natural light and soft indoor lighting can be used, and the light intensity is controlled within the range of 500-1000 lux. At the same time, the background should be simple, avoiding complex patterns or colors to reduce interference with face feature extraction.

[0086] Instruct the person being collected to maintain a natural and relaxed state, keep the head stable, and avoid large swings or overly exaggerated expressions. For each person being collected, collect multiple groups of videos, and the duration of each group of videos is 10-30 seconds to cover different instantaneous facial states. During the collection process, require the person being collected to perform some simple actions, such as natural blinking, slight eye rotation, smiling, etc. These actions help to obtain richer facial information.

[0087] Before collecting videos of different subjects to be collected, clean and calibrate the camera to ensure the consistency of image quality. At the same time, perform real-time preview and preliminary inspection on the collected videos. If blurring, occlusion or other quality problems are found in the videos, re-collect them immediately.

[0088] S102, Process the facial video through remote photoplethysmography to obtain a remote photoplethysmogram signal;

[0089] Specifically, extract image data frame by frame from the collected facial video. Use image analysis algorithms to identify facial skin areas, especially parts rich in blood vessels such as the cheeks and forehead. The skin color changes in these parts are related to blood volume changes, and photoplethysmography is based on this principle to detect the subtle fluctuations of blood volume changes.

[0090] According to the absorption and reflection characteristics of the skin to light of different wavelengths, select light of appropriate wavelengths for analysis. For example, green light (wavelength about 500 - 570nm) is usually used because the absorption difference of oxyhemoglobin and reduced hemoglobin in the blood by green light is relatively obvious, which can better reflect blood volume changes. By analyzing the light intensity changes of the skin area under green light irradiation in each frame of the image, a signal similar to a pulse wave is extracted.

[0091] Analyze the characteristic parameters of the extracted photoplethysmogram signal, such as the amplitude, frequency, waveform, etc. of the pulse wave. Different psychological and emotional states will cause changes in these parameters. For example, when a person is in a tense mood, the pulse wave frequency may increase and the amplitude may increase.

[0092] Combined with facial expression analysis, further determine psychological and emotional characteristics. Use techniques such as the Facial Action Coding System (FACS) to analyze the movements of facial muscles in the video, such as frowning, raising the corners of the mouth and other expressions, and combine this expression information with the characteristics of the pulse wave signal. For example, a frowning action accompanied by a specific change in the pulse wave frequency may imply anxiety. At the same time, observe the changes in pupil size, because the pupils also have different reactions under different emotions, such as the pupils may dilate when excited or frightened.

[0093] Perform light intensity analysis on each frame of the image in the facial video to obtain a signal of the change of light intensity over time, and extract the original rPPG signal based on the signal;

[0094] Preprocess the original rPPG signal to obtain a pure rPPG signal.

[0095] S103, Analyze the heart rate variability characteristics of the remote photoplethysmogram signal to obtain heart rate variability characteristics;

[0096] Specifically, heart rate variability features are extracted from the pure rPPG signal;

[0097] Among them, the heart rate variability features include time-domain feature parameters, frequency-domain feature parameters, and non-linear feature parameters.

[0098] Currently commonly used HRV indicators are as follows: (1) Time-domain indicators: Standard deviation of NN intervals (SDNN) is used to measure the overall magnitude of HRV and evaluate the overall regulation of the autonomic nervous system on heart rate; Root mean square of differences between adjacent NN intervals (RMSSD) is an indicator to measure the regulation of the vagus nerve on heart rate; Standard deviation of the average NN interval every 5 minutes (SDANN) mainly reflects sympathetic nerve tone; (2) Frequency-domain indicators: Very low frequency band (VLF, 0.003 - 0.04 Hz), low frequency band (LF, 0.04 - 0.15 Hz) reflect the modulation intensity of the sympathetic and parasympathetic nerves (mainly the sympathetic nerve), high frequency band (HF, 0.15 - 0.4 Hz) reflects the modulation intensity of the parasympathetic nerve, and the sympathetic / parasympathetic balance ratio (LF / HF) reflects the balance of the dual regulation of the sympathetic and parasympathetic nerves on the heart.

[0099] The sympathetic nerve activity of the total population with biased constitutions is reduced to a certain extent; the changing trends of the autonomic nerve activities of different biased constitutions are different. For example, the sympathetic / parasympathetic balance ratio of the yang-deficiency constitution is reduced; there is a certain linear correlation between the heart rate variability indicators and the biased constitution scores. A decision tree is a tree model that solves which value in which dimension is used for division to minimize the information entropy after division. The lower the information entropy, the more ordered it is, and the stronger the certainty of classification. According to certain rules, the data after HRV feature analysis is traversed cyclically starting from the parent node to find the currently decisive feature, and it is split into child nodes according to the value of this feature. The child nodes continue to be split according to the above rules as new parent nodes until the conditions are met or splitting is no longer possible. Therefore, choosing the appropriate root node is the key to splitting, and the judgment criterion is to calculate the impurity of the sample set, and the calculation method uses the information gain method. As the nodes of the decision tree are divided, its uncertainty should decrease. As the classification progresses, more and more information is known and the overall entropy will decrease. Suppose there are n categories in the sample set S, and the nine major types of TCM constitution classifications corresponding to n are peaceful constitution, qi-deficiency constitution, yang-deficiency constitution, yin-deficiency constitution, phlegm-dampness constitution, damp-heat constitution, blood stasis constitution, qi-stagnation constitution, and special endowment constitution.

[0100] The time-domain feature parameters include the deviation of heart rate variability from the average value, the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals, the root mean square of adjacent RR interval differences, the standard deviation of adjacent RR interval differences, the coefficient of variation of the standard deviation of RR intervals, the measurement value of HRV magnitude, and the degree of deviation of heart rate from the average value:

[0101] Calculate the deviation of heart rate variability from the average value through formula (1);

[0102]

[0103] In the formula, SDNN is the deviation of heart rate variability from the average value, N is the total number of R - R intervals, RR i is the i - th R - R interval, is the average value of the R - R intervals of N heartbeats;

[0104] Calculate the percentage of the number of adjacent R - R interval differences greater than 50 ms in the total number of R - R intervals through formula (2);

[0105]

[0106] In the formula, PNN50 is the percentage of the number of adjacent R - R interval differences greater than 50 ms in the total number of R - R intervals, NN50 is the number of adjacent R - R interval differences greater than 50 ms, and N is the total number of R - R intervals;

[0107] Calculate the root mean square of the differences between adjacent R - R intervals through formula (3);

[0108]

[0109] In the formula, RMSSD is the root mean square of the differences between adjacent R - R intervals, N is the total number of R intervals, R i+1 is the (i + 1)-th R - R interval;

[0110] Calculate the standard deviation of the differences between adjacent R - R intervals through formula (4);

[0111]

[0112] In the formula, RR′ i is the difference between the i - th adjacent R - R intervals;

[0113] Calculate the coefficient of variation of the standard deviation of R - R intervals through formula (5);

[0114]

[0115] In the formula, CVSD is the coefficient of variation of the standard deviation of R - R intervals, and RMSSD is the root mean square of the differences between adjacent R - R intervals;

[0116] Calculate the measurement value of HRV magnitude through formula (6);

[0117]

[0118] In the formula, CVnni is the measurement value of HRV magnitude;

[0119] Calculate the degree of heart rate deviation from the average value through formula (7);

[0120]

[0121] In the formula, std_hr is the degree of heart rate deviation from the average value, HR i represents the i-th previous heart rate, and HR is the heart rate.

[0122] The frequency domain characteristic parameter includes the ratio of low-frequency to high-frequency power spectra;

[0123] Calculate the ratio of low-frequency to high-frequency power spectra through formula (8);

[0124]

[0125] In the formula, LF is the low-frequency power spectrum, HF is the high-frequency power spectrum, f(λ) is the power spectral density function, and dλ is the small change in frequency.

[0126] The non-linear characteristic parameter includes short-term heart rate variability and long-term heart rate variability:

[0127] Calculate the short-term heart rate variability through formula (9):

[0128]

[0129] In the formula, SD1 is the short-term heart rate variability;

[0130] Calculate the long-term heart rate variability through formula (10):

[0131]

[0132] In the formula, SD2 is the long-term heart rate variability.

[0133] Calculate the sample entropy through formula (11);

[0134]

[0135] In the formula, SampEn(m, r) is the sample entropy, B m (r) is the number of vector pairs of length m whose distance is less than or equal to r, B m+1 (r) is the number of vector pairs of length m + 1 whose distance is less than or equal to r, is the limit when approaching infinity;

[0136] Calculate the information entropy through formula (12);

[0137]

[0138] Wherein, Gain(F) is the information gain of feature F, Info(S) is the information entropy of dataset S, and Info F (S) is the conditional entropy of dataset S under the condition of feature F.

[0139]

[0140] Wherein, p is the proportion of samples of different categories.

[0141]

[0142] Wherein, k is that sample S is divided into k parts.

[0143] S104. Construct an identification model, input the heart rate variability features into the identification model for training to obtain the optimal identification model;

[0144] Specifically, divide the heart rate variability features into a training set, a validation set, and a test set;

[0145] Train the identification model based on the training set;

[0146] Perform performance evaluation on the trained identification model based on the validation set to obtain an identification model that meets the performance conditions;

[0147] Evaluate the prediction results of the identification model that meets the performance conditions based on the test set to obtain the evaluation index corresponding to the identification model.

[0148] Perform performance evaluation on the trained identification model based on the validation set to obtain an identification model that meets the performance conditions; evaluate the similarity calculation results of the identification model that meets the performance conditions based on the test set to obtain the evaluation index corresponding to the identification model. Perform performance evaluation on the identification model to obtain a percentage score (that is, the highest score is 100 points and the lowest score is 0 points), and determine the identification model with a score greater than the set value based on the percentage score. For example, the identification model with a score greater than 90 points is an identification model that meets the performance conditions.

[0149] Calculate the evaluation index for the identification model that meets the performance conditions to obtain the evaluation index of the identification model, and calculate the evaluation value corresponding to each evaluation index. The evaluation value is used to represent the ability value of the identification model on the evaluation index.

[0150] S105. Input the heart rate variability features to be identified into the optimal identification model to obtain the identification results of the nine major types of traditional Chinese medicine constitutions corresponding to the heart rate variability features to be identified;

[0151] Specifically, the face video data of the user to be tested is obtained by using a camera, and the extracted HRV features are summarized according to the constitution and the mean value is calculated with the above features and indicators, so as to obtain the scores of nine constitution types under the HRV features, realize the classification and identification of traditional Chinese medicine constitutions, and carry out dialectical treatment.

[0152] Here, after the sample data to be tested is processed, standard data suitable for model input is obtained, and the classification category can be obtained through the above trained recognition model. For example, compared with the balanced constitution, the SDANN of the total biased constitution decreases (P < 0.05), indicating that the sympathetic nerve activity of the biased constitution population is lower than that of the balanced constitution population. The changing trends of the heart rate variability indexes of different biased constitutions are different: compared with the balanced constitution, the SDANN, VLF, and LF / HF of the yang-deficiency constitution population decrease (P < 0.05), and the HF increases (P < 0.05), that is, the decrease of the sympathetic-parasympathetic balance ratio LF / HF in the yang-deficiency constitution is caused by the combined effect of the decrease of sympathetic nerve activity and the increase of parasympathetic nerve activity; the SDANN and SDNN of the qi-deficiency constitution and blood stasis constitution populations are lower than those of the balanced constitution (P < 0.05), indicating that the sympathetic nerve activity of the qi-deficiency and blood stasis constitutions decreases, and the overall regulation effect of the autonomic nervous system on the heart rate weakens; the VLF of the phlegm-dampness constitution population decreases (P < 0.05), etc.

[0153] Figure 2 This is the system embodiment architecture diagram for non-contact identification of nine types of traditional Chinese medicine constitutions of the present invention; as Figure 2 shown, a system for non-contact identification of nine types of traditional Chinese medicine constitutions provided by an embodiment of the present invention includes the following steps:

[0154] The acquisition module 10 is used to acquire N face videos, where N is an integer greater than 10;

[0155] The preprocessing module 20 is used to process the face video by remote photoplethysmography to obtain the remote photoplethysmogram signal;

[0156] The feature analysis module 30 is used to perform heart rate variability feature analysis on the remote photoplethysmogram signal to obtain heart rate variability features

[0157] The construction module 40 is used to construct an identification model, input the heart rate variability features into the identification model for training, and obtain an optimal identification model;

[0158] The identification module 50 is used to input the heart rate variability features to be identified into the optimal identification model to obtain the traditional Chinese medicine constitution identification result corresponding to the heart rate variability features to be identified.

[0159] The preprocessing module 20 is further used for:

[0160] Perform light intensity analysis on each frame image in the facial video to obtain a signal of the change in light intensity over time, and extract the original rPPG signal based on the signal;

[0161] Preprocess the original rPPG signal to obtain a pure rPPG signal.

[0162] The feature analysis module 30 is further configured to:

[0163] Extract heart rate variability features from the pure rPPG signal;

[0164] Among them, the heart rate variability features include time domain feature parameters, frequency domain feature parameters, and non-linear feature parameters.

[0165] The time domain feature parameters include the deviation of heart rate variability from the average value, the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals, the root mean square of adjacent RR interval differences, the standard deviation of adjacent RR interval differences, the coefficient of variation of the standard deviation of RR intervals, the HRV magnitude measurement value, and the degree of deviation of heart rate from the average value:

[0166] Calculate the deviation of heart rate variability from the average value through formula (1);

[0167]

[0168] In the formula, SDNN is the deviation of heart rate variability from the average value, N is the total number of cardiac cycle intervals, RR i is the i-th RR interval, is the average value of the RR intervals of N heartbeats;

[0169] Calculate the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals through formula (2);

[0170]

[0171] In the formula, PNN50 is the percentage of the number of adjacent RR interval differences greater than 50 ms in the total number of RR intervals, NN50 is the number of adjacent RR interval differences greater than 50 ms, and N is the total number of RR intervals;

[0172] Calculate the root mean square of adjacent RR interval differences through formula (3);

[0173]

[0174] In the formula, RMSSD is the root mean square of adjacent RR interval differences, N is the total number of R intervals, RR i+1 is the (i + 1)-th RR interval;

[0175] Calculate the standard deviation of the difference between adjacent RR intervals through formula (4);

[0176]

[0177] In the formula, RR′ i is the difference between the i-th adjacent RR intervals;

[0178] Calculate the coefficient of variation of the standard deviation of RR intervals through formula (5);

[0179]

[0180] In the formula, CVSD is the coefficient of variation of the standard deviation of RR intervals, and RMSSD is the root mean square of the differences between adjacent RR intervals;

[0181] Calculate the measurement value of HRV magnitude through formula (6);

[0182]

[0183] In the formula, CVnni is the measurement value of HRV magnitude;

[0184] Calculate the degree of deviation of heart rate from the average value through formula (7);

[0185]

[0186] In the formula, std_hr is the degree of deviation of heart rate from the average value, and HR i represents the i-th previous heart rate, and HR is the heart rate.

[0187] The frequency domain characteristic parameters include the ratio of low-frequency to high-frequency power spectra;

[0188] Calculate the ratio of low-frequency to high-frequency power spectra through formula (8);

[0189]

[0190] In the formula, LF is the low-frequency power spectrum, HF is the high-frequency power spectrum, f(λ) is the power spectral density function, and dλ is the small change in frequency.

[0191] The non-linear characteristic parameters include short-term heart rate variability and long-term heart rate variability:

[0192] Calculate short-term heart rate variability through formula (9);

[0193]

[0194] In the formula, SD1 is the short-term heart rate variability;

[0195] Calculate long-term heart rate variability through formula (10);

[0196]

[0197] Wherein, SD2 is the long-term heart rate variability.

[0198] Calculate the sample entropy through formula (11);

[0199]

[0200]

[0201] Wherein, SampEn(m, r) is the sample entropy, and B m (r) is the number of vector pairs of length m whose distance is less than or equal to r, and B m+1 (r) is the number of vector pairs of length m + 1 whose distance is less than or equal to r, is the limit when approaching infinity;

[0202] Calculate the information entropy through formula (12);

[0203] Gain(F) = Info(S) - Info F (S) Formula (12);

[0204] Wherein, Gain(F) is the information gain of feature F, Info(S) is the information entropy of dataset S, and Info F (S) is the conditional entropy of dataset S under the condition of feature F.

[0205] Figure 3 This is the schematic diagram of the physical structure of the electronic device provided by the embodiment of the present invention. As Figure 3 shown, the electronic device 60 includes: a processor 601 (processor), a memory 602 (memory), and a bus 603;

[0206] Among them, the processor 601 and the memory 602 complete communication with each other through the bus 603;

[0207] The processor 601 is used to call the program instructions in the memory 602 to execute the methods provided by the above method embodiments, for example, including: collecting N face videos, where N is an integer greater than 10; processing the face videos through remote photoplethysmography to obtain the remote photoplethysmogram signal; performing heart rate variability feature analysis on the remote photoplethysmogram signal to obtain heart rate variability features; constructing an identification model, inputting the heart rate variability features into the identification model for training to obtain an optimal identification model; inputting the heart rate variability features to be identified into the optimal identification model to obtain the identification results of the nine traditional Chinese medicine physical types corresponding to the heart rate variability features to be identified.

[0208] This embodiment provides a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the methods provided in the above method embodiments. For example, it includes: collecting N face videos, where N is an integer greater than 10; processing the face videos through remote photoplethysmography to obtain the remote photoplethysmogram signals; performing heart rate variability feature analysis on the remote photoplethysmogram signals to obtain heart rate variability features; constructing an identification model, inputting the heart rate variability features into the identification model for training to obtain an optimal identification model; inputting the heart rate variability features to be identified into the optimal identification model to obtain the identification results of the nine major traditional Chinese medicine constitution types corresponding to the heart rate variability features to be identified.

[0209] Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: ROM, RAM, magnetic disk, optical disk, or other storage media that can store program codes.

[0210] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

[0211] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0212] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A non-contact method for identifying nine types of constitutions in traditional Chinese medicine, characterized in that: The non-contact method for identifying the nine types of constitutions in traditional Chinese medicine specifically includes: Collecting N number of human face videos, where N is an integer greater than 10; Processing the facial video of the human face by remote photoplethysmography to obtain the remote photoplethysmographic signal; Performing heart rate variability characteristic analysis on the remote photoplethysmography signal to obtain a heart rate variability characteristic; Constructing a recognition model, inputting the heart rate variability characteristics into the recognition model for training, and obtaining an optimal recognition model; The heart rate variability feature to be identified is input into the optimal identification model to obtain the identification results of the nine types of constitutions in traditional Chinese medicine corresponding to the heart rate variability feature to be identified.

2. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 1, characterized in that: The step of performing heart rate variability characteristic analysis on the remote photoplethysmography signal to obtain the heart rate variability characteristic includes: Performing light intensity analysis on each frame of the face video to obtain a signal of light intensity variation over time, and extracting an original rPPG signal based on the signal; The original rPPG signal is preprocessed to obtain a pure rPPG signal.

3. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 2, characterized in that: The step of performing heart rate variability characteristic analysis on the remote photoplethysmography signal to obtain the heart rate variability characteristic further comprises: Extracting heart rate variability features from the pure rPPG signal; Wherein, the heart rate variability characteristics include time domain characteristic parameters, frequency domain characteristic parameters and nonlinear characteristic parameters.

4. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 3, characterized in that: The time domain characteristic parameters include the deviation of heart rate variability from the mean value, the percentage of adjacent RR intervals with a difference greater than 50ms in the total number of RR intervals, the root mean square of adjacent RR interval differences, the standard deviation of adjacent RR interval differences, the coefficient of variation of the RR interval standard deviation, the HRV magnitude measurement value and the degree of heart rate deviation from the mean value: The heart rate variability deviation from the mean is calculated by formula (1); Where SDNN is the deviation of heart rate variability from the mean, N is the total number of heart beat intervals, and RR is i is the i-th RR interval, is the average value of the RR interval of N heart beats; The percentage of the number of adjacent RR intervals with a difference greater than 50 ms to the total number of RR intervals was calculated using formula (2); Where, PNN50 is the percentage of the number of adjacent RR intervals with a difference greater than 50ms to the total number of RR intervals, NN50 is the number of adjacent RR intervals with a difference greater than 50ms, and N is the total number of RR intervals; The root mean square of the difference between adjacent RR intervals was calculated by formula (3); Where RMSSD is the root mean square of the difference between adjacent RR intervals, N is the total number of R intervals, and RR i+1 is the i+1th RR interval; The standard deviation of the difference between adjacent RR intervals was calculated by formula (4); In the formula, RR′ i is the difference between the i-th adjacent RR intervals; The coefficient of variation of the standard deviation of the RR interval was calculated by formula (5); Where, CVSD is the coefficient of variation of the standard deviation of the RR interval, and RMSSD is the root mean square of the difference between adjacent RR intervals; The HRV magnitude measurement value is calculated by formula (6); In the formula, CVnni is the HRV size measurement value; The degree to which the heart rate deviates from the average value is calculated by formula (7); Where std_hr is the degree to which the heart rate deviates from the average value, HR i represents the heart rate in the i-th previous period, and HR is the heart rate.

5. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 3, characterized in that: The frequency domain characteristic parameters include the ratio of low-frequency to high-frequency power spectra; The ratio of low-frequency to high-frequency power spectra is calculated by formula (8); Where LF is the low-frequency power spectrum, HF is the high-frequency power spectrum, f(λ) is the power spectrum density function, and dλ is the small change in frequency.

6. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 3, characterized in that: The nonlinear characteristic parameters include short-term heart rate variability and long-term heart rate variability: The short-term heart rate variability is calculated by formula (9): Where SD1 is short-term heart rate variability; Long-term heart rate variability is calculated using formula (10): Where SD2 is the long-term heart rate variability.

7. The method for non-contact identification of nine types of constitution in traditional Chinese medicine according to claim 1, characterized in that: The non-contact method for identifying nine types of constitutions in traditional Chinese medicine also includes: The sample entropy is calculated by formula (11); In the formula, SampEn(m, r) is the sample entropy, B m (r) is the number of vector pairs with a length of m whose distance is less than or equal to r, B m+1 (r) is the number of vector pairs of length m+1 whose distance is less than or equal to r, is the limit when it tends to infinity; The information entropy is calculated by formula (12); Gain(F)=Info(S)-Info F (S) Formula (12); Where Gain(F) is the information gain of feature F, Info(S) is the information entropy of data set S, and Info F (S) is the conditional entropy of the data set S under the condition of feature F.

8. A system for non-contact identification of nine types of constitution in traditional Chinese medicine, characterized in that: include: A collection module, used for collecting N number of human face videos, where N is an integer greater than 10; A preprocessing module, used for processing the face video of the human face by remote photoplethysmography to obtain the remote photoplethysmography signal; A feature analysis module is used to perform heart rate variability feature analysis on the remote photoelectric volume pulse wave signal to obtain a heart rate variability feature. A construction module is used to construct a recognition model, input the heart rate variability characteristics into the recognition model for training, and obtain an optimal recognition model; The identification module is used to input the heart rate variability characteristics to be identified into the optimal identification model to obtain the TCM constitution identification result corresponding to the heart rate variability characteristics to be identified.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A non-transitory computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images